A dataset generated for illustration of the principal stratification analysis. This dataset represents the common case of non-compliance.
sim_data_Cox## `sim_data_Cox` A data frame with 1,000 rows and 7 columns:
Principal Strata: "never taker", "complier" or "always taker"
Randomized treatment arm: 0 = control, 1 = treatment
Actual treatment arm: 0 = control, 1 = treatment
True outcome: event time
Censor time
Event indicator. 1 means true outcome is observed; 0 means otherwise
The observed event time or censor time
The dataset represents the scenario where actual treatment might not be in compliance with the randomized (assigned) treatment. Defiers and always-takers are ruled out, leaving two strata, "never-taker" and "complier" randomly sampled with probability 0.3, 0.7 respectively. The assigned treatment \(Z\) is randomized with 0.5 probability for either arm. The true event time \(T\) is given by the following Weibull-Cox distribution
\(Y \sim Weibull-Cox(theta = 1, mu = 0.3)\)
\(Y \sim Weibull-Cox(theta = 1, mu = 2 - 0.6*Z)\)
and the censor time \(C\) is uniformly drawn between 0.5 and 2.
The exclusion restriction assumption holds for never-takers in this generated dataset.